Using TMPGEnc project files in CCE. Which does what?

[Letricheur]

I have been using a TMPGEnc *.tpr project file created from a DVD2AVI *.d2v, wrapped in VFAPI, loaded into CCE 2.62 to create non-compliant 480X480 MPEG-2s for SVCD (actually XSVCD). So far, it works,.... slowly. What I'd like to know is, what has TMPGEnc contributed to the encoding? I basically set up TMPGEnc as though I am going to encode the *.d2v using TMPGEnc the way I used to. I set the output to interlaced, 480X480, 4:3 aspect (same as the DVD in question), choose CQ-VBR, max 4000 bitrate, high search motion, etc. I save the project file, close down TMPGEnc, change the *.tpr to *.avi through VFAPI and load into CCE. There I leave it 4:3, use DCT 10, zig-zag, NOT progressive, linear quant settings and run a CBR at 4000 bitrate. Next I run it through again as 3-pass VBR, range 1500-4000, average 3500. The output is certainly 480x480 4:3 and everything else I ordered in CCE (checked with bitrate viewer).

So...Which of the original TMPGEnc settings were 'used' by CCE? The only one I can say for certain was 480x480 resizing. Also, this seems to be slower (0.282 RT) than using avisynth (0.33, and which I am now having trouble with, hence the change to TMPGEnc)

I suppose I am just curious but also want to know which settings are best to use in TMPGEnc?

[Kedirekin]

Basically, when you use a TMpg tpr file for frame serving via VFAPI, TMpg acts as a filter engine - very similar to AviSynth. As you've already realized, resize happens in Tmpg. If you enable any of the filters on the advanced tab (such as IVTC), they'll be invoked as well. None of the encoding settings are used though - it truly does act as a frame server.

The reason it is slower is that VFAPI always operates in the RGB color space. This means that, as DVD2AVI decodes the vobs, it has to perform a YUV-to-RGB conversion to comply with VFAPI. Then CCE has to perform an RGB-toYUV conversion during encoding. These extra color space conversion slow down the whole process.

AviSynth on the other hand does all its processing in the YUV color space (at least, if you use mpeg2source), thereby avoiding the time consuming color space conversions.

[Letricheur]

Thanks for the reply, very clear. I think I'll have to try avisynth once again.